Hierarchical Federated Unlearning for Large Language Models

Yisheng Zhong, Zhengbang Yang, Zhuangdi Zhu
October 2025
PDF

Abstract

Large Language Models (LLMs) are increasingly integrated into real-world applications, raising concerns about privacy, security and the need to remove undesirable knowledge. We propose a federated unlearning approach for LLMs that is scalable and privacy preserving, with task-specific adapter learning and hierarchical merging.

Type
Preprint
Publication
KDD Workshop on Federated Learning for Data Mining and Graph Analytics, 2025
Federated Unlearning LLMs Privacy
Zhuangdi Zhu
Zhuangdi Zhu
Assistant Professor (Tenure-Track)

My research centers around accountable, scalable, and trustworthy AI, e.g., decentralized machine learning, knowledge transfer for supervised and reinforcement learning, debiased representation learning, etc.

Related